On the robustness of exoplanet atmospheric detections: insights from extensive simulations

Abstract

The classical picture of our Solar System being the archetypal outcome of planet formation has been rendered obsolete by the astonishing diversity of extrasolar-system architectures. From rare hot-Jupiters to abundant super-Earths and sub-Neptunes, most detected exoplanets have no analogs in our system, and their interior and atmospheric compositions remain largely unknown. Fortunately, new methodologies enable us to analyze exoplanet atmospheres, inferring their compositions, temperatures, dynamics, and even formation pathways. Specifically, ground-based high-resolution Doppler spectroscopy (HRDS) can disentangle spectral-line profiles of weak exo-atmospheric signals from the dominating features of Earth's atmosphere in the observed flux. For over a decade, HRDS has focused on hot Jupiters (close-orbiting gas giants) due to their high signal-to-noise ratio, which makes them ideal laboratories for advancing our knowledge. However, there have been concerns regarding potential biases in exo-atmospheric-detection methods, hindering comparative planetology. Here we propose a modeling framework based on extensive simulations of HRDS exo-atmospheric observations to systematically explore in-silico underlying biases in commonly-used pipelines, particularly under the presence of observational noise. Our findings show that exo-atmospheric detection-significances are highly contingent on details of the analysis-pipeline used, with different techniques responding differently to noise: A given technique may fail to recover a true-signal that is detected by another. Noise effects in the computed significances are non-trivial and pipeline-dependent. Statistical analyses provide a complementary tool to contextualize signal-significances, which will gain in relevance as we move towards studying the atmospheres of smaller, potentially habitable exoplanets with even weaker signals.